Is testing determinism somehow easier?
We've learned quite a bit about the capabilities of deterministic systems, and some of their limitations. That's the subject matter of computing theory.
We can and do test whether deterministic models are adequate to explain observations. That's how we learned that certain quantum events cannot be deterministic and can only be modeled as random instead. It's also how we learned that chaotic or complex dynamic systems can be impossible to predict even though they are deterministic.
Recent developments in AI demonstrate that it's possible for a deterministic system to generate a stream of verbal narrative and/or selected actions from processed memory and present inputs. And also that such a system can work despite being unable to correctly account for why it produces the outputs it does. This doesn't prove or even strongly suggest that such AIs perceive themselves the way we do or perceive anything at all, but it's consistent with the possibility that we're in that same boat.